Informatics Publishing Limited

Naresh Kumar Nagwani ¹, Shrish Verma ²

Affiliations
1 Computer Science & Engineering, National Institute of Technology Raipur
2 Electronics & Tel. Communication Engg. National Institute of Technology Raipur

Abstract

Sampling is an important and necessary step in mining large size databases and is also very useful in performing mining operations, where performance is a critical issue. This study focuses on identifying the effect of sample size in classification of software bugs. To analyze the effect of sample size, experiments are performed using a number of classification algorithms with varities of sample sizes using the software bug repositories of three large open source software's namely Android, Mozilla and MySql. The relationship between the sample size with two primary classification performance parameters accuracy and F-measure is explored in this study. From experiments, it is identified that the parameter F-measure is affected more by the sample size than accuracy.

Keywords

Sampling, Sample Size, Classification, Software Bug, Performance, Classifier Evaluation

Full Text

   

Sudhakar Pandey ¹, Naresh Kumar Nagwani ², Chandan Kumar ³

Affiliations
1 Department of IT, NIT Raipur - 492010, IN
2 Department of CSE, NIT Raipur - 492010, IN
3 Department of IT, NIT Raipur - 492010,, IN

Abstract

There are many facets of Wireless Sensor Networks (WSN), which provoke research interest. One of the major being routing protocol, i.e. the task of transmitting data from source node to sink node. SPIN (Sensor Protocols for Information via Negotiation) being one of them, which efficiently disseminates information among sensor node in an energy-constrained wireless sensor network. Although in time constrained operation like war zone it is not suited because it take much more time to deliver information among sensor node, also it consume more energy to broadcast data among sensor node that results energy drain faster and node dead quickly. Hence we proposed a modified version of SPIN routing protocol with network being formed using clustering technique, called Cluster Based SPIN (CB-SPIN) routing protocol. For simulation of both protocols SPIN and CB-SPIN, MATLAB platform is used and result show that CB-SPIN gives better performance in terms of time and energy.

Keywords

Cluster, Energy, Lifetime, Routing Protocol, Sensor Node, SPIN, WSN

Full Text

   

Naresh Kumar Nagwani ¹, Shrish Verma ¹

Affiliations
1 National Institute of Technology, Raipur-492001, CG, IN

Abstract

Software bug classification is one of the important and popular problems in software engineering. Recently number of algorithms and techniques are presented to automate this process. Software bug data contains number of attributes like bug-id, summary (title), description, comments, status, version etc. Most of the important attributes holds text data. Lingo and STC (Suffix Tree Clustering) both are popular text clustering algorithms used in web mining. In this paper Lingo and STC algorithms are used to classify the software bugs. Classification using clustering methodology is used to create the software bug classes from software bug clusters. In this methodology first clusters are created and then appropriate labels are assigned to the clusters, which indicate the class label for the clusters. Both of these algorithms Lingo and STC are implemented as the part of Carrot2 framework. The software bug repository data is integrated and passed to Carrot2 framework for applying Lingo and STC algorithms. Lingo and STC algorithms are compared for software bug classification task. The comparison is done using various clustering parameters: the number of clusters generated, purity of the clusters and entropy of the clusters created etc.

Keywords

Software Bug Classification, Lingo Clustering, STC Clustering, Software Bug Clustering, Software Bug Repository.

Full Text

     

Naresh Kumar Nagwani ¹, Shrish Verma ¹

Affiliations
1 National Institute of Technology, Raipur-492001, CG, IN

Abstract

Association rule mining is one of very popular data mining method and number of organizations uses this technique to find the frequent item-sets of products to improve the benefits of organizations. There are number of available algorithms for association rule mining which takes multiple scans of database. The complexities of association rule algorithms primarily depends on number of database scan, so by reducing the number of database scans one can improve the time complexity of these algorithms. The purpose of this proposed algorithm is to reduce the number of database scans for discovering the temporal association rules by applying P-Tree algorithm for temporal association rules which takes just one scan of database to find out the association rules.
A new pattern tree algorithm for mining temporal association rules in databases is introduced. This algorithm uses P-Tree (Pattern-Trees) structures for finding temporal association rules in databases. According to different time periods associated with transactions in temporal databases, it will initiate the number of P-Trees and according to time information in transactions it inserts the transactions in created appropriate trees, then using P-Tree association rule mining algorithm it finds out the frequent sets in this P-Tree and then these frequent items are merged with different time periods which will give the association rules with valid time periods. The proposed algorithm is divided in two phases in first phase all item within the transactions are inserted in different P-Trees on which the frequent item-sets are taken out and in second (merge phase) these frequent items are merged and time associates with these items are in listed which indicates that these frequent items are frequent in this time periods. Algorithm is implemented in C++ under Linux platform and evaluated results are compared with existing popular algorithm PPM (Progressive Partition Miner) for discovering temporal association rule.

Keywords

Temporal Association Rules, Temporal Data Mining, P-Tree, Incremental Data Mining.

Full Text

     

Debasish Sena ¹, Naresh Kumar Nagwani ¹

Affiliations
1 National Institute of Technology, Raipur 492 010, IN

Abstract

India is one of the fast developing countries in the world with a growth rate of 6.4%. Rapid industrialization is the main cause behind such growth. Although industrialization is of utmost importance for growth, sustainability of ecology is also a matter of concern. India has a vast coastline, but the saline water is not suitable for industrialization; so groundwater is the primary source for both industrialization and human consumption. Agriculture plays a major role in India's economy and irrigation is also dependent on groundwater to some extent. Hence the study of groundwater levels is the need of the hour. In this study, time-series techniques like fuzzy time-series analysis and ARIMA are utilized for forecasting monthly groundwater levels. Experiments are performed on the datasets collected from different regions of India. The experimental results demonstrate that fuzzy time series analysis yields more accurate forecast of groundwater levels compared to the ARIMA model. The results of this study can be utilized for planning a suitable policy for groundwater use and its proper regulation to avoid future crisis.

Keywords

Fuzzy Logic, Groundwater Level, Prediction Models, Time-Series Forecasting.

Full Text